Legionnaires' disease (LD) is the second most common cause of bacterial pneumonia in the U.S., accounting for 2-9% of community-acquired pneumonia cases (Cunha, Burillo, & Bouza, 2016; Stout & Yu, 1997). Developed countries around the globe have experienced an increase in LD incidence since the 2000s (Beaute, Zucs, de Jong, & European Legionnaires' Disease Surveillance Network, 2013; Centers for Disease Control & Prevention [CDC], 2011; Farnham, Alleyne, Cimini, & Balter, 2014). From 2000-2014 in the U.S., legionellosis incidence, which includes LD and the milder, less commonly reported Pontiac fever, increased 286%, from 0.42-1.62 annual cases per 100,000 people (Garrison et al., 2016). This trend persists even after age-adjustment (CDC, 2011).
The majority of cases reported in the U.S. and worldwide occur sporadically with no identified source (Che et al., 2008; Fields, Benson, & Besser, 2002). The most common sources are speculated to be home potable water, travel-associated potable water, and evaporative cooling towers (Bhopal, 1995; Ricketts, Joseph, Lee, & Wilkinson, 2012). Through spatial analysis of LD in England and Wales, Ricketts and coauthors (2012) estimated that 20% of sporadic cases could be attributed to cooling towers.
Transmission of LD occurs through inhalation or aspiration of water containing Legionella. Legionella is a waterborne pathogen found in most aqueous environments and proliferates in warm, stagnant water. Legionella bacteria commonly inhabit amoeba as intracellular parasites and thrive in biofilms formed on surfaces (Cunha et al., 2016). Conditions for proliferation are commonly found in evaporative cooling towers. Prevalence of the bacteria in these structures has ranged from 2-87% and variations exist likely due to sample selection, maintenance practices, and possibly local cooling tower regulations (Lau, Maqsood, Harte, Caughley, & Deacon, 2013; Mouchtouri, Goutziana, Kremastinou, & Hadjichristodoulou, 2010; Ragull et al., 2007; Turetgen, Sungur, & Cotuk, 2005; Witherell et al., 1986).
Both large and small LD community outbreaks have been caused by cooling towers. A 2014 review article described 19 outbreaks attributable to cooling towers with case counts ranging from 7-449 cases and a 6.3% average case fatality rate (Walser et al., 2014). A hotel cooling tower in the South Bronx neighborhood of New York City caused a 2015 outbreak that sickened 138 people and killed 16. Clinical Legionella isolates matched the strain of Legionella found in the cooling tower (Weiss et al., 2017). In response to this outbreak, both New York City and the state of New York issued emergency regulations requiring cooling tower registration, inspection, and Legionella testing (New York State Public Health and Health Planning Council and the Commissioner of Health, 2015).
The highest incidence of legionellosis in the U.S. occurs in the Mid-Atlantic Region. Allegheny County, Pennsylvania, which is part of this Mid-Atlantic Region, experiences rates 4 times higher than the U.S. age-adjusted rate (Allegheny County Health Department, 2014). Over two thirds of LD cases reported annually in Allegheny County are of unknown origin. These cases are unrelated to outbreaks or healthcare facilities. Cooling tower-related LD has not been identified recently in Allegheny County, but has occurred in the past. Investigating the conditions of cooling towers is an important component of LD prevention, especially in an area with a high burden of the disease. The purpose of this survey was to assess Legionella prevalence in Allegheny County cooling towers and then identify areas of improvement for cooling tower maintenance and Legionella contamination prevention in Allegheny County.
Cooling Tower Maintenance Survey
Buildings selected for the survey included those that house populations who are susceptible to LD. These buildings included hospitals, skilled nursing facilities, assisted living facilities, personal care homes, and senior apartment buildings identified through the Pennsylvania Department of Health and Department Human Services. Allegheny County senior apartment buildings were identified through a Google search using search terms "senior apartment AND Allegheny county." City- and county-owned buildings in Allegheny County, Pennsylvania, were also surveyed and identified through the Allegheny County Housing Authority, the Housing Authority of the City of Pittsburgh, and the Allegheny County Facilities Management Department.
A questionnaire was completed over the phone or sent via e-mail or fax based on facility preference. The questionnaire began with vetting questions to ensure the most knowledgeable persons at the facility completed the survey (see supplemental document). Survey questions were based on guidelines from the American Society of Heating, Refrigerating, and Air-Conditioning Engineers (ASHRAE), the Cooling Technology Institute, U.S. Environmental Protection Agency (2016), and the World Health Organization.
Structural questions addressed building size, number of cooling towers, cooling tower location, and name of water authority. Maintenance questions addressed use of water treatment professional, cooling tower cleaning and inspection procedures, water filtration, basin emptying, biocide treatment and monitoring, record keeping, bacterial load testing, and Legionella testing. Finally, facilities were asked by ACHD to consent to testing their cooling tower basin water for Legionella.
Cooling Tower Sampling
At consenting facilities, ACHD staff selected a single, random cooling tower for testing if the facility had multiple cooling towers. The cooling tower's make, model, serial number, year installed, and size (tonnage) were recorded. Basin water temperature was measured using a digital probe thermometer and pH was measured using test strips. Free and total chlorine were measured using test strips (range 0-10 ppm at increments of 0, 0.5, 1, 2, 4, and 10 ppm).
Basin water was collected in sterile 125 mL plastic bottles. Bottles were filled to 30 mL with basin water and a drop of sterile 0.1 N sodium thiosulfate was added to the bottle immediately after water collection using a sterile, disposable transfer pipette. Water samples were sent to the ACHD Public Health Laboratory on the same day as sample collection. Water samples were stored at 5 [degrees]C until processing.
Water samples were cultured for Legionella pneumophila (Lp) within four days of collection at the ACHD Laboratory. Each specimen was plated onto GVPC agar directly after acid treatment and heat treatment. Specifically, Legionella acid buffer was added to each sample for 15 min at room temperature. Samples were heat treated at 50 [degrees]C for 30 min in a water bath before plating. Plates were incubated at 35 [degrees]C and read at 3 and 7 days. Any identified colonies were picked and plated on SBA and GVPC agar and incubated overnight at 35 [degrees]C. Isolates that grew on GVPC agar were tested with Oxoid Legionella Latex Test kit (Oxoid Ltd.) and confirmed positive for Lp serogroups 1, 3, 5, 6, Poly 1-14, or b-m with a direct fluorescent antibody test (Monoclonal Technologies, Inc.; rabbit anti-Legionella IgG fluorescein labeled).
Whole Genome Sequencing and Phylogenetics
Genomic DNA was extracted at the University of Pittsburgh School of Medicine, Infectious Disease Epidemiology Research Unit, using the Qiagen DNA easy Tissue Kit on a QIAcube according to manufacturer's instructions (Qiagen). The DNA was eluted in 10 mM Tris/1 mm EDTA and sequenced according to the method of Baym and coauthors (2005) using Illumina Nextera genomic libraries on a MiSeq v2 (500-cycle) kit.
Fastq Reads were trimmed and assembled using SPAdes version 3.9.0. Assemblies were annotated using Prokka version 0.1.1. The sequencing depth ranged from 36-94X. The assemblies had a median of 96 contigs per sample with an average assembly length of 3.7 Mbp and an average N50 of 200,000 bp.
Sequence types (ST) were identified using SRST2. Reads were aligned to reference assembly, LEG551, using BWA-MEM version 0.7.12-r1039. For ST2329 pairwise comparisons, LEG443 was used as the reference genome. Single-nucleotide polymorphisms (SNPs) were identified using GATK HaplotypeCaller version 3.5 with a ploidy of 1. SNPs with low mapping quality (MQ 60.0), low variant confidence (QD